Skip to contentSkip to conversation

Subject: [PATCH] SQUASHME: pnfsd: nfs4pnfsd.c should dprint under NFSDDBG_PNFS

PATCH Boaz Harrosh 3 participants 9 messages Nov 9, 2011 – Nov 14, 2011

Subsystems (2)
NFS client NFS server
lore
Share & tools
lore

Boaz Harrosh

author
#0 lore raw

I've been squinting at pnfsd dprints for ages and aching under the wait. So I headed up to the nfsd/debug.h to add a PNFS channel, and what do you know? there is one! It's used by nfs4pnfsds.c. Surly nfs4pnfsd.c is more _PNFS than _PROC.

While at it I changed nfs4pnfsdlm.c to, also already defined, NFSDDBG_FILELAYOUT which was unused before this patch.

CC Steve Dickson <[email protected]>
Signed-off-by Boaz Harrosh <[email protected]>
---
 fs/nfsd/nfs4pnfsd.c   |    2 +-
 fs/nfsd/nfs4pnfsdlm.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
fs/nfsd/nfs4pnfsd.c +1 -1
index 18b8fca..f514ebe 100644
@@ -23,7 +23,7 @@
 
 #include "pnfsd.h"
 
-#define NFSDDBG_FACILITY                NFSDDBG_PROC
+#define NFSDDBG_FACILITY                NFSDDBG_PNFS
 
 /* Globals */
 static u32 current_layoutid = 1;
fs/nfsd/nfs4pnfsdlm.c +1 -1
index abc4d83..0f0441a 100644
@@ -30,7 +30,7 @@
 #include "nfsfh.h"
 #include "nfsd.h"
 
-#define NFSDDBG_FACILITY                NFSDDBG_PROC
+#define NFSDDBG_FACILITY                NFSDDBG_FILELAYOUT
 
 /* Just use a linked list. Do not expect more than 32 dlm_device_entries
  * the first implementation will just use one device per cluster file system
Signature · 1 line
1.7.6.2
#1 lore raw
Quoted 4 lines · Boaz Harrosh

I've been squinting at pnfsd dprints for ages and aching under the wait. So I headed up to the nfsd/debug.h to add a PNFS channel, and what do you know? there is one! It's used by nfs4pnfsds.c. Surly nfs4pnfsd.c is more _PNFS than _PROC.

Hey now, nfs4pnfsd.c has enough problems of its own without being accused of surliness.

--b.

Quoted 45 lines

While at it I changed nfs4pnfsdlm.c to, also already defined, NFSDDBG_FILELAYOUT which was unused before this patch.

CC Steve Dickson <[email protected]>
Signed-off-by Boaz Harrosh <[email protected]>
---
 fs/nfsd/nfs4pnfsd.c   |    2 +-
 fs/nfsd/nfs4pnfsdlm.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
fs/nfsd/nfs4pnfsd.c +1 -1
index 18b8fca..f514ebe 100644
@@ -23,7 +23,7 @@
 
 #include "pnfsd.h"
 
-#define NFSDDBG_FACILITY                NFSDDBG_PROC
+#define NFSDDBG_FACILITY                NFSDDBG_PNFS
 
 /* Globals */
 static u32 current_layoutid = 1;
fs/nfsd/nfs4pnfsdlm.c +1 -1
index abc4d83..0f0441a 100644
@@ -30,7 +30,7 @@
 #include "nfsfh.h"
 #include "nfsd.h"
 
-#define NFSDDBG_FACILITY                NFSDDBG_PROC
+#define NFSDDBG_FACILITY                NFSDDBG_FILELAYOUT
 
 /* Just use a linked list. Do not expect more than 32 dlm_device_entries
  * the first implementation will just use one device per cluster file system
Signature · 7 lines
1.7.6.2


--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
#8 lore raw
In reply to Boaz Harrosh
Quoted 10 lines · Boaz Harrosh

I've been squinting at pnfsd dprints for ages and aching under the wait. So I headed up to the nfsd/debug.h to add a PNFS channel, and what do you know? there is one! It's used by nfs4pnfsds.c. Surly nfs4pnfsd.c is more _PNFS than _PROC.

While at it I changed nfs4pnfsdlm.c to, also already defined, NFSDDBG_FILELAYOUT which was unused before this patch.

CC Steve Dickson <[email protected]>
Signed-off-by Boaz Harrosh <[email protected]>

Makes sense. Thanks!

Benny

Quoted 31 lines
---
 fs/nfsd/nfs4pnfsd.c   |    2 +-
 fs/nfsd/nfs4pnfsdlm.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
fs/nfsd/nfs4pnfsd.c +1 -1
index 18b8fca..f514ebe 100644
@@ -23,7 +23,7 @@
 
 #include "pnfsd.h"
 
-#define NFSDDBG_FACILITY                NFSDDBG_PROC
+#define NFSDDBG_FACILITY                NFSDDBG_PNFS
 
 /* Globals */
 static u32 current_layoutid = 1;
fs/nfsd/nfs4pnfsdlm.c +1 -1
index abc4d83..0f0441a 100644
@@ -30,7 +30,7 @@
 #include "nfsfh.h"
 #include "nfsd.h"
 
-#define NFSDDBG_FACILITY                NFSDDBG_PROC
+#define NFSDDBG_FACILITY                NFSDDBG_FILELAYOUT
 
 /* Just use a linked list. Do not expect more than 32 dlm_device_entries
  * the first implementation will just use one device per cluster file system
#2 lore raw
In reply to J. Bruce Fields
Quoted 8 lines · J. Bruce Fields
Quoted 4 lines · Boaz Harrosh

I've been squinting at pnfsd dprints for ages and aching under the wait. So I headed up to the nfsd/debug.h to add a PNFS channel, and what do you know? there is one! It's used by nfs4pnfsds.c. Surly nfs4pnfsd.c is more _PNFS than _PROC.

Hey now, nfs4pnfsd.c has enough problems of its own without being accused of surliness.

:-)

Quoted 52 lines

--b.

Quoted 45 lines

While at it I changed nfs4pnfsdlm.c to, also already defined, NFSDDBG_FILELAYOUT which was unused before this patch.

CC Steve Dickson <[email protected]>
Signed-off-by Boaz Harrosh <[email protected]>
---
 fs/nfsd/nfs4pnfsd.c   |    2 +-
 fs/nfsd/nfs4pnfsdlm.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
fs/nfsd/nfs4pnfsd.c +1 -1
index 18b8fca..f514ebe 100644
@@ -23,7 +23,7 @@
 
 #include "pnfsd.h"
 
-#define NFSDDBG_FACILITY                NFSDDBG_PROC
+#define NFSDDBG_FACILITY                NFSDDBG_PNFS
 
 /* Globals */
 static u32 current_layoutid = 1;
fs/nfsd/nfs4pnfsdlm.c +1 -1
index abc4d83..0f0441a 100644
@@ -30,7 +30,7 @@
 #include "nfsfh.h"
 #include "nfsd.h"
 
-#define NFSDDBG_FACILITY                NFSDDBG_PROC
+#define NFSDDBG_FACILITY                NFSDDBG_FILELAYOUT
 
 /* Just use a linked list. Do not expect more than 32 dlm_device_entries
  * the first implementation will just use one device per cluster file system
Signature · 7 lines
1.7.6.2


--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

author
#3 lore raw
In reply to J. Bruce Fields
Quoted 9 lines · J. Bruce Fields
Quoted 4 lines · Boaz Harrosh

I've been squinting at pnfsd dprints for ages and aching under the wait. So I headed up to the nfsd/debug.h to add a PNFS channel, and what do you know? there is one! It's used by nfs4pnfsds.c. Surly nfs4pnfsd.c is more _PNFS than _PROC.

Hey now, nfs4pnfsd.c has enough problems of its own without being accused of surliness.

Allow me to disagree! the nfs4pnfsd.c is a very nice pretty clean, well formatted and well implemented pNFS-Server implementation. Believe me I have seen a few other pNFS and Parallel servers implementation and this is surgery room clean compare to the other. (If you look at the point before pnfs-exp and spNFS patches)

The only real mess in there is the mess already inherited from nfsd, like the great and grate messy locking. Though it's better then the rest of the NFSD code.

So you see, I do think it could be accused of "surliness"

Quoted 3 lines

--b.

Quoted 1 line

--Be

author
#4 lore raw
Quoted 16 lines · Boaz Harrosh
Quoted 9 lines · J. Bruce Fields
Quoted 4 lines · Boaz Harrosh

I've been squinting at pnfsd dprints for ages and aching under the wait. So I headed up to the nfsd/debug.h to add a PNFS channel, and what do you know? there is one! It's used by nfs4pnfsds.c. Surly nfs4pnfsd.c is more _PNFS than _PROC.

Hey now, nfs4pnfsd.c has enough problems of its own without being accused of surliness.

Allow me to disagree! the nfs4pnfsd.c is a very nice pretty clean, well formatted and well implemented pNFS-Server implementation. Believe me I have seen a few other pNFS and Parallel servers implementation and this is surgery room clean compare to the other. (If you look at the point before pnfs-exp and spNFS patches)

Quoted 5 lines

The only real mess in there is the mess already inherited from nfsd, like the great and grate messy locking. Though it's better then the rest of the NFSD code.

So you see, I do think it could be accused of "surliness"

O, by the way I did not mean there is not lots of places for improvements and that the code is perfect.

If you see things you don't like, or buggy or, could be done better, please tell us so we can fix them.

Quoted 9 lines
Quoted 3 lines

--b.

Quoted 1 line

--Be
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#5 lore raw
In reply to Boaz Harrosh
Quoted 10 lines · Boaz Harrosh
Quoted 3 lines · J. Bruce Fields

Hey now, nfs4pnfsd.c has enough problems of its own without being accused of surliness.

Allow me to disagree! the nfs4pnfsd.c is a very nice pretty clean, well formatted and well implemented pNFS-Server implementation. Believe me I have seen a few other pNFS and Parallel servers implementation and this is surgery room clean compare to the other. (If you look at the point before pnfs-exp and spNFS patches)

OK, I hope so! And we *are* getting closer to finishing off the last of the 4.1 stuff (though with just me it's not going to happen for 3.3.), so I hope to be looking at that soon.

Quoted 2 lines

The only real mess in there is the mess already inherited from nfsd, like the great and grate messy locking.

You mean the overuse of the state lock, or something else?

--b.

Quoted 1 line

Though it's better then the rest of the NFSD code.

author
#6 lore raw
Quoted 4 lines · J. Bruce Fields

On Thu, Nov 10, 2011 at 08:04:21AM -0800, Boaz Harrosh wrote:

You mean the overuse of the state lock, or something else?

Yes and the places that causes to unlock / lock because of recursion or bad lock ordering.

So it's not as clean as it could be

Thanks
Boaz

Quoted 3 lines

--b.

Quoted 1 line

Though it's better then the rest of the NFSD code.

#7 lore raw
Quoted 8 lines · Boaz Harrosh
Quoted 4 lines · J. Bruce Fields

On Thu, Nov 10, 2011 at 08:04:21AM -0800, Boaz Harrosh wrote:

You mean the overuse of the state lock, or something else?

Yes and the places that causes to unlock / lock because of recursion or bad lock ordering.

That's purely a pNFS problem, right?

--b.